Skip to main content

​
Participant Types

Participant types define the structure for endpoints sharing LLM access within rooms.

​
ParticipantInfo

Complete participant information including endpoint, model, specs, and status. 
import { z } from "zod";

const ParticipantInfo = z.object({
  id: z.string(),
  nickname: z.string(),
  model: z.string(),
  endpoint: z.string().url(),
  config: GenerationConfig,
  specs: MachineSpecs,
  status: ParticipantStatus,
  joinedAt: z.number(),
  lastSeen: z.number(),
});

type ParticipantInfo = z.infer<typeof ParticipantInfo>;
​
id
string
required
Unique identifier for the participant
​
nickname
string
required
Display name chosen by the participant
​
model
string
required
Name of the LLM model being shared (e.g., “llama3:8b”, “gpt-4”)
​
endpoint
string
required
OpenAI-compatible API endpoint URL (Ollama, LM Studio, etc.)
​
config
GenerationConfig
required
Generation parameters for this participant’s model. See GenerationConfig.
​
specs
MachineSpecs
required
Hardware specifications of the participant’s machine. See MachineSpecs.
​
status
ParticipantStatus
required
Current availability status: "online", "busy", or "offline"
​
joinedAt
number
required
Unix timestamp (milliseconds) when the participant joined the room
​
lastSeen
number
required
Unix timestamp (milliseconds) of the last health check

​
ParticipantStatus

Enum representing the availability status of a participant. 
const ParticipantStatus = z.enum(["online", "busy", "offline"]);

type ParticipantStatus = z.infer<typeof ParticipantStatus>;
  • "online" - Participant is available and can accept requests
  • "busy" - Participant is currently processing a request
  • "offline" - Participant has not sent a health check within the timeout period

​
MachineSpecs

Hardware specifications for a participant’s machine. 
const MachineSpecs = z.object({
  gpu: z.string().optional(),
  vram: z.number().optional(),
  ram: z.number().optional(),
  cpu: z.string().optional(),
});

type MachineSpecs = z.infer<typeof MachineSpecs>;
​
gpu
string
GPU model (e.g., “NVIDIA RTX 4090”)
​
vram
number
Video RAM in GB
​
ram
number
System RAM in GB
​
cpu
string
CPU model (e.g., “AMD Ryzen 9 7950X”)

​
GenerationConfig

Generation parameters compatible with OpenAI-like APIs. These are the common parameters supported by most providers. 
const GenerationConfig = z.object({
  temperature: z.number().min(0).max(2).optional(),
  top_p: z.number().min(0).max(1).optional(),
  max_tokens: z.number().optional(),
  stop: z.array(z.string()).optional(),
  frequency_penalty: z.number().min(-2).max(2).optional(),
  presence_penalty: z.number().min(-2).max(2).optional(),
  seed: z.number().optional(),
});

type GenerationConfig = z.infer<typeof GenerationConfig>;
​
temperature
number
Sampling temperature between 0 and 2. Higher values make output more random.
​
top_p
number
Nucleus sampling parameter between 0 and 1
​
max_tokens
number
Maximum number of tokens to generate
​
stop
string[]
Array of sequences where generation will stop
​
frequency_penalty
number
Penalty for token frequency between -2 and 2
​
presence_penalty
number
Penalty for token presence between -2 and 2
​
seed
number
Random seed for deterministic generation

​
Health Check Constants

Constants that control participant health monitoring. 
// Health check interval in milliseconds
export const HEALTH_CHECK_INTERVAL = 10_000;

// Time after which a participant is considered offline (3 missed health checks)
export const PARTICIPANT_TIMEOUT = HEALTH_CHECK_INTERVAL * 3; // 30_000
​
HEALTH_CHECK_INTERVAL
number
default:"10000"
Interval in milliseconds between health checks (10 seconds)
​
PARTICIPANT_TIMEOUT
number
default:"30000"
Timeout in milliseconds before marking a participant as offline (30 seconds, or 3 missed health checks)

​
Usage Example

import {
  ParticipantInfo,
  ParticipantStatus,
  MachineSpecs,
  GenerationConfig,
} from "@gambiarra/core";

const participant: ParticipantInfo = {
  id: "participant_abc123",
  nickname: "Alice's MacBook",
  model: "llama3:8b",
  endpoint: "http://localhost:11434/v1",
  config: {
    temperature: 0.7,
    max_tokens: 2048,
  },
  specs: {
    gpu: "Apple M2 Max",
    vram: 32,
    ram: 64,
    cpu: "Apple M2 Max",
  },
  status: "online",
  joinedAt: Date.now(),
  lastSeen: Date.now(),
};

// Check if participant is available
if (participant.status === "online") {
  console.log(`${participant.nickname} is available`);
}

Build docs developers (and LLMs) love

Get started for freeTalk to us